Environmental Impact Assessment of a hypothetical accident during ship transport of spent nuclear fuel from Andreeva Bay to FSUE ATOMFLOT

(1)

Environmental Impact Assessment of a hypothetical accident during ship transport of spent nuclear fuel from Andreeva Bay to FSUE

ATOMFLOT

DSA Report

Dec 2021, number 5

(2)

Referanse

Brown JE, Hosseini A, Dowdall, M, Amundsen, I.

Environmental Impact Assessment of Spent Nuclear Fuel transported from Andreeva Bay to FSUE ATOMFLOT.

DSA-rapport 2021:5. Østerås, Direktoratet for strålevern og atomsikkerhet, 2021.

Emneord

Andreeva Bay, Spent Nuclear Fuel, Transport, Environmental Impact Assessment

Resymé

Vurdering av potensielle miljøkonsekvenser av en ulykke under skipstransport av brukt atombrensel fra Andrejevbukta.

Reference

Brown JE, Hosseini A, Dowdall, M, Amundsen, I.

Environmental Impact Assessment of Spent Nuclear Fuel transported from Andreeva Bay to FSUE ATOMFLOT.

DSA Report 2021:5. Østerås: Norwegian Radiation and Nuclear Safety Authority, 2021.

Language: Norwegian.

Key words

Andreeva Bay, Spent Nuclear Fuel, Transport Abstract

Assessment of potential environmental

consequences of an accident during transport of spent nuclear fuel by sea from Andreeva Bay.

Godkjent:

Ingar Amundsen, fung. avdelingsdirektør, avdeling kunnskapsutvikling og internasjonal atomsikkerhet

Publisert 2021-12-21

Sider 56

DSA,

Postboks 329, 0213 Oslo Norge.

Telefon 67 16 25 00

Faks 67 14 74 07

Email [email protected]

dsa.no

ISSN 2535-7339

(3)

DSA-rapport Nummer 5 Publisert des 2021

Environmental Impact Assessment of a hypothetical accident

during ship transport of spent nuclear fuel from Andreeva Bay

to FSUE ATOMFLOT

(4)

1 Background

The Andreeva Bay facility of the Northwestern Centre for Radioactive Waste Management ‘SevRAO’ and FSUE ATOMFLOT are both located on the Kola Peninsula. Andreeva Bay, located on the northwestern shore of the Kola Peninsula, lies some 45 km from the Russian-Norwegian border. FSUE ATOMFLOT is located in the Kola Bay on the northern periphery of Murmansk. Andreeva Bay was constructed and commissioned between 1961-1963 and functioned as a service base for nuclear submarines for some 25 years and also received, treated and stored (on a temporary basis) solid and liquid radioactive wastes (SRW and LRW). The facility stopped receiving spent nuclear fuel (SNF) and wastes in 1989. The current

temporary SNF and RW storage facility at Andreeva Bay is managed by the Northwestern Centre for radioactive waste management SevRAO, branch of FSUE RosRAO (NWC SevRAO - branch of FSUE RosRAO). The facility is host to the SNF of approximately 100 nuclear submarine reactors and large amounts of radioactive waste, all stored under poor conditions. The Andreeva Bay site is being remediated and the most significant phase of the decommissioning is the removal of SNF from its current storage and its transport for processing.

The process of removal of SNF from the Andreeva Bay site, technical details regarding the SNF and its handling and the environmental implications of a possible release of activity during these operations have been previously assessed by Shilov et al. (2017) in a detailed assessment. This assessment included detailed information as to preventative and mitigatory systems and procedures put in place to minimize risk of and impact from accidental releases of radioactive materials. One aspect of the removal process that was not covered in great detail was the actual transport of the SNF by sea to FSUE ATOMFLOT. The passage to FSUE ATOMFLOT is a journey of approximately 100 km and takes 6 hours. The SNF is carried onboard the specially built, Italian funded nuclear fuel carrier Rossita, which has two isolated holds with a capacity of 720 tonnes. The SNF will be transported in transport casks of the designation TUK-108/1 and TK- 18 whith capacities of 49 spent fuel assemblies (SFA). ТUК-108/1 (ТUК-18) (type B (U) accord with relevant international standards for such containers and meet all requirements as outlined by the

International Atomic Energy Agency (IAEA) as well as those outlined by relevant Russian regulatory bodies.

This report concerns itself with the potential environmental consequences of a release of activity during the transport of SNF by sea from Andreeva Bay to the facilities of FSUE ATOMFLOT, elements of which are funded by the Norwegian Governments Nuclear Action Plan.

(6)

2 Methodology

2.1 Source Term

For the estimation of a source term, reference was made to Shilov et al. (2017) which provides an impact assessment of various stages of the process for the removal of spent nuclear fuel from the Andreeva Bay facility. In the description of beyond design-based accidents, indicative figures are provided for the release fractions of ¹³⁷Cs and ⁹⁰Sr under various scenarios and the corresponding activity. Combination of the values facilitates a rough estimate of the activity of ¹³⁷Cs and ⁹⁰Sr contained within the 49 assemblies housed within 1 TUK container. Based on these values, the amount of ¹³⁷Cs and ⁹⁰Sr contained within 1 TUK was estimated as being 2.2 x 10¹⁴ Bq and 1.8 x 10¹⁴ Bq respectively.

For the purposes of simplicity and conservatism, consequences were estimated for the release of the entire contents one TUK cask. Adoption of these values has no bearing on estimating the probability of any particular accident type or release and served solely to facilitate estimation of consequences for a given amount of activity released. The amount should be considered as extremely conservative.

2.2 Scenario Description

For the purposes of the work reported upon here, hypothetical releases were considered as having occurred at each of three open water locations: the first at the mouth of Zapadnaya Litsa, the second midway between Andreeva Bay and Murmansk and the third at the mouth of the Kola Bay. It should be considered that all three are so close together that there is little or no difference oceanographically, from the perspective of the applied marine contaminant dispersion model, between the selected locations.

Release points in the immediate vicinity of Andreeva Bay and FSUE ATOMFLOT were not chosen for two reasons; there were model limitations for near-shore release points and previous impact assessments have dealt with releases while loading and unloading at Andreeva Bay and FSUE ATOMFLOT respectively.

2.3 Modelling Methodology

2.3.1 Advection and dispersion of radionuclides in the marine environment

The circulation of the Barents Sea in the southwestern parts is characterized by a general motion of waters from west to east. Inflow from the Norwegian Sea takes place through the Bear Island Channel.

Close to the coast, low salinity (S = 34.4) water of the Norwegian Coastal Current (NCC) carries runoff from the Baltic and the Norwegian coasts into the southern Barents Sea (Fig. 1, green, Fig. 2). It continues eastward as the Murmansk Coastal Current. Additional low salinity water is added as it passes the entrance to the White Sea and the mouth of the Pechora River, so the salinity remains low (S = 34.6). Most of this

‘‘coastal’’ water passes into the Kara Sea through the Kara Gate. River runoff and net precipitation are small, and the NCC is the major freshwater source for the Barents Sea (Tantsiura, 1959; Loeng, 1991). In addition, warm and salty Atlantic derived water of the Norwegian Atlantic Current in the Nordic Seas enters north of the coastal current and south of Bear Island (Fig. 1, red). On its way to the east, the warm water of Atlantic origin loses heat and several varieties of water densities are formed, part of these sink to the bottom of the Barents Seas. Approaching Novaya Zemlya, the Atlantic derived waters turn north while cooling down further. To the north, a strong front separates the Atlantic derived water masses from the cold and fresh waters stemming from the Arctic. These move southwestward and leave the Barents Sea between Svalbard and around Bear Island (Fig. 1, blue). The coastal currents are strongly density driven

(7)

offshore of the coast of Murmansk will drift east, and northeast. However, in particular in summer months, the wind field patterns in the Barents Sea may induce westward motion also in the southern part for a few weeks (Harms and Karcher, 2005). Figure 3 shows a timeseries of typical surface (0-20m) velocities in the coastal current 50-100 km offshore from the Murrmansk coast as simulated by the NAOSIM model (see section III for a model description). The only extended period of westward flow in this area occurred in summer 1998, which was characterized by a specific weather pattern which suppressed the eastward flow and even led to westward return flow over larger areas in the southern Barents Sea and parts of the Kara Sea (e.g. Harms and Karcher, 2005). Further north, in the central Barents Sea and around the Central Bank (Fig. 1), currents are much more variable and easily can involve westward motion.

Figure 1: The Barents Sea. Red arrows show Atlantic water currents, blue arrows Arctic currents and green arrows currents of coastal waters (from Eriksen et al., 2017).

(8)

The numerical model used for the current set of experiments is a version of NAOSIM (North Atlantic/Arctic Figure 2: Spatial distribution of permanent surface density currents in the Barents Sea according to the results of

diagnostic calculations (Terziev et al., 1990). Figures in circles are the serial numbers of currents. Figures without circles indicate velocity in cm/s (Kagan et al., 2015). Franz Josef Land Current (1), East Spitsbergen Current (2), Suidcap Current (3), Medvezhinskoe Current (4), Persey (Southwestern) Current (5), Central Current (6), Litke Current (7), warm permanent Southern Spitsbergen Current (8), Nordkap Current (9) (which is divided into the northern (10), central (11), and southern (12) branches), Murmansk Current (13), Murmansk coastal Current (14), Canin Current (15), Kolguev Pechora Current (16), Novaya Zemlya Current (17), discharge surface currents Belomorskoe (18) and Pechora (19).

Figure 3: Simulated surface velocities (positive southeastward, along shore) (0-20m) about 60 km offshore of the coast near Murmansk, in the Murmansk coastal current (red dot in Fig. 2; NAOSIM, monthly means).

(9)

Geophysical Fluid Dynamics Laboratory modular ocean model MOM-2 (Pacanowski, 1995) and incorporates a dynamic-thermodynamic sea ice model with a viscous-plastic rheology (Hibler, 1979). The version used here has 30 unevenly spaced levels in the vertical, starting from 20 m thickness down to 100 m depth with the thickness gradually increasing with depth. The model domain covers the Nordic Seas, the Arctic Ocean and the northern North Atlantic down to about 50ºN. The Canadian Archipelago allows throughflow

between the central Arctic and the Labrador Sea. The model has an open boundary in Bering Strait where a total inflow for the barotropic mode of 0.8 Sv is imposed to cover approximate amount of observed inflow of Pacific Water into the Arctic Ocean. It also has an open boundary in the south, where barotropic flow from a larger scale version of the model is prescribed. At both boundaries the flow profiles can develop freely. The open boundary conditions have been implemented following Stevens (1991), thereby allowing the outflow of tracers and the radiation of waves. The initial hydrography in January 1948 is adopted from the PHC winter climatology (Steele et al., 2001), while a yearly mean climatology is used as a reference for surface salinity restoring on a time scale of 180 days. The restoring of sea surface salinity is a common method used to prevent the ocean salinity from drastically drifting away from the observed ocean state (Steele et al., 2001). Sea surface salinity restoring compensates for a mismatch between freshwater forcing data (e.g., precipitation and runoff) and model physics. In the absence of restoring the model drift is arbitrary, depending on the combination of forcing data set and model physics. Parametrization of river runoff is employed using negative salt fluxes proportional to seasonal climatologies of runoff for each of the major rivers which follows the AOMIP protocol (Holloway et al., 2007). The model is driven with daily atmospheric forcing from 1948 to 2013 (NCEP/NCAR reanalysis(Kalnay et al., 1996])). NAOSIM has been used successfully in a number of applications focusing on Northern Sea circulation (Karcher et al. 2008, Gerdes et al. 2005) and tracer dispersion (Karcher et al. 2004; 2012). It also has been used for studies identifying worst case locations for radioactive releases in the Nordic Seas (Brown et al., 2016a; Kauker et al., 2016) and contributing to an environmental impact analysis of a potential release form the Russian submarine K-27 (Hosseini, et al., 2017; Karcher et al., 2017).

This work reported upon here encompass marine dispersion of potentially released soluble radionuclides arising offshore from a hypothetical, modelled source of radioactive substances transported from

Andreeva Bay to Murmansk, both areas being close to the Norwegian border. Sediment interaction has not been considered; the half-life of relevant tracers has been assumed to be much longer than the modelled dispersion period. The experiments performed encompass a continuous release of 1 TBq/year into the surface layer (0-20m) over the time periods:

A) January 1998 – January 2001 B) January 2007 – January 2010 C) January 2010 – January 2013

The period A January 1998 to January 2001 was chosen because of the exceptional duration of westward flow in the first year. Periods B and C were chosen as more realistic periods from the recent past, with conditions in the Barents Sea more typical for the last decade. These periods also exhibit short events of westward flow. Data was stored as monthly means of the tracer concentration in the common NetCDF format. A web-based database management tool, developed by OASys, was prepared to facilitate interrogation of the data based on geographical and temporal information for each of the experiments.

2.3.2 Selection of input data for use in the subsequent environmental impact assessment A point was chosen in the marine environment that would yield the most elevated activity concentrations

(10)

management tool as described elsewhere (Hosseini et al., 2016). Surface concentrations of ¹³⁷Cs and ⁹⁰Sr in seawater were selected for an area encompassing the most elevated levels associated with the main plume of contamination for the release scenarios. While these data might correspond to a single grid cell in the modelling domain, the spatial averaging involved was nonetheless quite substantial covering an area of approximately 900 km². A spatial averaging of this magnitude might over-compensate for the

migratory/peripatetic nature of the animal and fish species covered (Sazykina, 1998 and Wienerroither et al., 2011). However, no higher resolution was available for the model. This observation provided the rationale for considering the surface water only (to a depth of 10 m) in the model output, and for which activity concentrations were at a maximum, to characterize more robustly the potential levels in the environment that could be attained for the release. For consistency, and to address similar concerns regarding spatial averaging, surface water data were also used for the human assessment.

2.3.3 Atmospheric-terrestrial advection dispersion modelling

For the estimation of the atmospheric transport of radionuclides to and deposition over Norway, a decision was made to use earlier simulations performed as part of an environmental risk assessment involving the analysis of radionuclide release and dispersion from the Russian nuclear submarine K-159 (Hosseini et al.

2017). This decision was based on considerations related to the proximity of the two locations (Andreeva Bay and sinking site of the submarine) and the fact that, for any new set of simulations, changes regarding prevailing wind-fields, leading to worst case radionuclide deposition conditions in Norway, would be expected to be minor. A regional weather model could hence be suitably adopted from previous simulations undertaken for K-159.

The model used for simulation of the atmospheric dispersion is called SNAP (Severe Nuclear Accident Program). SNAP is a Lagrangian particle model which was developed at the Norwegian Meteorological Institute, MET (Bartnicki et al., 2011) to simulate atmospheric dispersion of radioactive debris in an emergency situation. The SNAP model has been run with a generic source term twice a day for a period between 1980 and 2012, using the meteorological database NORA10-EI (Reistad et al., 2011). This resulted in a database of depositions containing more than 24 000 deposition maps over northern Europe with a resolution of approx. 11 km. One additional argument for reusing these deposition maps has been the low resolution of these maps. In Fig. 4, the current position of the submarine and the envisaged route of Rosita is shown. However, with regards to Norway a worst-case scenario would be an accident at Andreeva Bay, which is much closer to the Norwegian border. The difference in proximity of release points has been accounted for by conducting some SNAP exploratory simulations, based on release from the Andreeva Bay location, and their comparison with previous results for K-159. The cursory investigation showed that the difference with regards to deposition in Norway, as a result of Andreeva Bay being closer, would be at maximum 3 times the value associated with an accident at the sinking site of the K-159 submarine. Hence, a factor of 3 has been applied to ensure that conservatism is preserved in the reported estimations.

(11)

From the obtained deposition maps, only those with deposition over the Norwegian mainland were analysed, and the cases with the largest total deposition to Norway have been selected for more in-depth investigation. For these calculations a possible accident scenario considering a release of 220, 180 and 26 TBq of ¹³⁷Cs, ⁹⁰Sr and ⁸⁵Kr, respectively, to the atmosphere is assumed. The release height is up to 100 m and it was assumed to be short – one hour, in the form of radioactive particles. Based on these runs, a worst case was identified, and maximum depositions and time integrated air concentrations were estimated.

2.3.4 Modelling transfer of radionuclides in food-chains

For the purpose of deriving activity concentrations in representative plants and animals and for human foodstuffs, food-chain models have been applied. For convenience, these have been split into marine (Appendix A) and terrestrial models (Appendix B).

The suite of equations describing the systems (as described below) has been constructed within the modelling platform software ECOLEGO-6. ECOLEGO is a simulation software tool that is used for creating dynamic models and performing deterministic and probabilistic simulations. Further details are reported in Avila et al. (2005).

Result of simulations from marine dispersion models, as described above in Section 2.3, have been provided in the form of a continuous release of 1 TBq/y over a 3-year period. Our outputs from the model in Bq/l (for a generic radionuclide), therefore, correspond to this unit release rate and need to be scaled accordingly.

To make the employed source term inputs compatible with the way the hydrodynamic model was set up, it was assumed that the activity was released at a constant and continuous rate over a 3-year period. A

‘scaling factor’ was thus derived, by dividing the total release of a given radionuclide (as specified

previously) by 3 years. This scaling factor was then applied to every determination of activity concentration for the same radionuclide for every time point. The transfer of contaminants to marine plants and animals is often estimated using concentration ratios (or factors) as described elsewhere (IAEA, 2004; IAEA, 2014).

Figure 4: The current position of the submarine K-159 and the envisaged route of Rosita from Andreeva Bay to Murmansk.

The circle in the enlarged part has a radius of about 11 km which is equal to resolution of the deposition maps used.

(12)

are prevalent as oppose to conditions under which environmental activity concentrations are changing rapidly with time. In these latter cases, dynamic models are more appropriate for determining biological transfer (Vives i Batlle et al., 2008). For this reason, marine food-chain models were applied (Appendix A) to estimate activity concentrations of ⁹⁰Sr and ¹³⁷Cs in representative aquatic species.

A terrestrial food-chain model was used to provide input for both the derivation of ingestion doses for humans and for the assessment of doses to wild plants and animals. The approach follows closely the methodologies adopted in quantifying the impact of hypothetical releases arising from the sunken nuclear submarine K-159, the details of which can be found elsewhere (Hosseini et al., 2017).

In view of available data, it was most appropriate to split the modeling into flora (wild grass/grasses, herbs and shrub) and fauna (deer/herbivorous mammal and rat/burrowing mammal) partly based on the

classifications given in UNSCEAR (2008). More details regarding the approach taken and modelling works conducted to account for the transfer of contamination along the considered terrestrial food chains can be found in Appendix B.

2.3.5 Doses to humans

Once activity concentrations are derived, models are required to derive dose estimates for both a representative person and representative plants and animals. Standard methodologies were used for the calculation of human exposures (see e.g. IAEA, 2003) from various exposure pathways such as:

→ Ingestion from contaminated foodstuffs

→ Inhalation of radionuclides

→ Exposure arising from a passing plume of contamination - cloud-shine

→ Exposure from contaminated soil or shore sediments

The equations expressing the total annual effective dose from the above pathways is given in Appendix C.

The human dose assessment has been spilt into 2 main parts – the first relating to the local region in proximity to the release location for the Andreeva accident scenario and the second for Norwegian territory.

For a representative person on the Kola Peninsula, it was assumed that ingestion of marine products constitutes the dominant exposure pathway. In earlier work and in this regard, IASAP (International Arctic Seas Assessment Project – IAEA, 1998) considered a representative of the average local Russian

population (using local marine food products). This group was located on the Kola Peninsula. The dietary habits postulated for these populations were:

→ Sea fish, 50 kg/a (assumed to have been caught in the Barents Sea)

→ Molluscs, 0.5 kg/a

→ Crustaceans, 1 kg/a

The same dietary data has been used for the calculations presented in this report. The activity

concentrations of ¹³⁷Cs and ⁹⁰Sr in marine biota have been derived using the marine food chain models described in Appendix A using time series radionuclide activity concentrations in seawater as described above.

For a representative person in the European context and with respect to marine pathway transfer, a high rate consumer of seafood was considered. The EFSA (European Food Safety Authority) Comprehensive European Food Consumption Database (EFSA, 2011) has been used as a source of dietary information. It is

(13)

a high percentile consumer from the statistics based on the entire country. Consumption rates were provided for fish and seafood of 63 g/d and 250 g/d for an average person and person in the 99^th

percentile, respectively (Hosseini et al., 2017). The 99^th percentile has been selected for subsequent dose calculations. Since no differentiation is made between fish and seafood in statistical data for the dietary information, the assumed breakdown has been taken as 50 % fish and 50% crustacean and molluscs (the latter forming equal components of their 50 %). This may seem somewhat arbitrary and differs from Hosseini et al. (2017) where the diet was assumed to consist of 100 % fish but allows for the inclusion of marine foodstuffs, in the dose calculation, that may be potentially more contaminated than fish (cf. activity concentrations of radionuclides in molluscs and fish in various figures in this report). For consumers of fish and seafood, fish tends to form the predominant dietary component as evidenced by the Kola Peninsula representative person (see above and IAEA, 1998) and as such our breakdown of the diet has been selected to reflect this observation. The activity concentrations in seafood have been obtained via the kinetic models described in Appendix A.

In view of uncertainty regarding where the marine organisms have been prior to capture, a conservative assumption was made that they were caught in a small area corresponding to locations where radionuclide activity concentrations were at a maximum. The time series data for radionuclide activity concentrations in seawater for this area, from 3-dimensional hydrodynamic modelling, has been taken as input to food-chain transfer models. The output from the modelling work is in the form of activity concentrations in organisms with time. For the sake of conservatism, the maximum values from these data series were used in human dose calculations.

In considering exposure due to consumption of terrestrial foodstuffs, the representative person was considered as consumer of natural products in high deposition areas in Finnmark exposed to initial plume and deposition For the sake of conservatism, the maximum values from model simulations for the time integrated activity concentrations of radionuclides in air over Norwegian territory (in Finnmark) were used for inhalation and cloud-shine dose estimates applying the equations given in Appendix C (Equations C-2 and C-3). Exposure from contaminated soil was derived using the maximum deposition levels of

radionuclides in Finnmark County. For this, it was conservatively assumed that a person is found in continual contact with contaminated soil over a 1-year period. The reality, even in the absence of orders to shelter or evacuate, would be that individuals are shielded from ground contamination for prolonged periods as people spend a substantial fraction of time indoors. Finally, ingestion doses for people have been derived using various assumptions concerning diet.

The kinetic model described in Appendix B was used where possible to derive activity concentrations in various foodstuffs. The maximum deposition levels of radionuclides in Finnmark County have been used as input to the model and appropriate activity concentration from the time series data generated in this process are then used for subsequent ingestion dose calculations. Further details concerning the use of dietary surveys in the assessment and underlying assumptions can be found in Hosseini et al. (2017).

2.3.6 Doses to biota

Once activity concentrations in environmental media and biota were collated and calculated, dose-rates were derived through application of the ERICA Tool (Brown et al., 2008). A selection was made of representative organisms in line with the guidance given in (ICRP, 2008). For our study, sea mammal (seals), seabirds, crustaceans, molluscs and fish were considered for the marine environment whereas a large mammal, small burrowing mammal and shrub were considered for the terrestrial environment. The basic underlying equations utilize activity concentration data in order to derive internal (Dint) and external (Dext) absorbed dose-rates (in units of µGy/h). The total absorbed dose-rate is the sum of these

(14)

3 Results

3.1 Advection- dispersion of radionuclides in the environment

3.1.1 Marine dispersion

As expected, release of the soluble radionuclide starting in January leads to a steady dispersion of the contaminant to the east, spreading a bit southeastward towards the Russian coast and to the north, for all three experiments (Fig. 5, 7, 9). The major pathways are developing into a split plume with one branch entering the Kara Sea via the Kara Strait, and a second branch moving northeast along the coastline of Novaya Zemlya (Fig. 5, 7, 9). For all experiments, once the dispersed contaminant has reached the frontal zone to the Arctic waters in the northern part of the Barents Sea, part of the contaminant is mixed into these Arctic waters and is carried to the west towards Svalbard. Three years after the start of the release, the tongue of contaminated waters reaching Svalbard is evident in all three experiments (Fig. 5, 7, 9). In the central Barents Sea, some westward spread of the contaminant occurs at an earlier stage, a few months after start of the release, driven by the westward flow. However, the short duration of these westward drift periods are not long enough to allow dispersion far into Norwegian waters. Concentrations stay low, even for the case of the extended period of westward flow in 1998, concentrations above 10^-6 Bq/m³ barely cross the 30ºE longitude line (Fig. 5a).

The situation for the southern Barents Sea, just westward to the release, is somewhat different. Here the front between contaminated and uncontaminated water is strongest, the large spatial gradients leading to large differences in the contamination by just 50 km difference in location. A timeseries of surface water concentration (0-20 m) close to Vardø reveals concentrations up to 0.8 x 10^-2 Bq/m³ in the late summer of 1998, a few months after start of the release (Fig. 6). When it comes to local details of the flow we must ask for caution with the results of the present model. Local flow features not resolved by the model, plus tidal mixing in the area, may lead to higher concentrations than those simulated here, in close proximity to the front. Timeseries of the concentration in the surface waters (0-20m) for the central Barents Sea near 73ºN 30ºE (Fig. 6b) show the arrival of the westward drift signal from 1998 with 0.5 10^-6 Bq/m³ also in late summer. In contrast, the temporal development on the southeastern tip of Svalbard, near Edgeøya (Fig. 6) visualizes a pulse-like arrival of concentrations reaching the maximum 0.3 x 10^-6 Bq/m³ three years after start of the release. However, concentrations might keep on rising in the period after the simulation stopped.

For the period January 2007 to January 2010, peak concentrations close to Vardø reach 0.2 x 10^-2 Bq/m³ for one month in summer 2007 (Fig. 8a). For this period the peak concentrations in the central Barents Sea near 73ºN 30ºE (Fig. 8b) occur in summer 2008. In late 2009, near the end of the integration, the highest concentration close to Edgeøya (Fig. 8c) is about 0.2 x 10^-5 Bq/m³, again at the very end of the 3-year integration, so we might expect a continued period of contamination.

For the latest release period, January 2010 to January 2013, we find a rather regular pulse of increased concentrations each summer near Vardø, highest in the first summer of 2010, while increasing

concentrations occur each of the three summers in the central Barents Sea (Fig. 10b). For Svalbard (Edgeøya) (Fig. 10c) the situation is similar to the other periods, with a maximum of the concentrations at the end of the simulation.

(15)

Figure 5: Experiment A: 1/1998-1/2001, Surface (0-20m) concentrations in September 1998 (top) and January 2001 (bottom) after continuous release of 1 TBq/y off Murmansk in 10^-2 Bq/m^3.

(16)

Figure 6: Concentration of the radionuclide (10^-2 Bq/m³) in surface water (0-20 m) for Experiment A (1/1998 – 1/2001), just offshore of Vardø.(top), southern Barents Sea, near 73ºN 30ºE (middle) and at the southeastern tip of Svalbard near Edgeøya (bottom).

(17)

As considered in more detail above, it appears that in all cases, i.e. three simulation periods, the dominant flow, and thus dispersion, happens to be East-Northeast, with periods of a few weeks up to a few months of westward flow in the southern and central Barents Sea. This leads to a strong front between contaminated and uncontaminated waters on the western flank of the dispersion plume, which is located very close to, or moving a short distance beyond, the Norwegian-Russian border. Thus, highest concentrations in

Norwegian regions can be expected near Vardø, when the front moves westward for some weeks to months. Further north in the Barents Sea, when the contaminated plume reaches beyond the front to the Arctic waters which move west-southwest, contaminant loaded water is carried towards Svalbard. This happens in year 3 after the release has started.

Figure 7: Experiment B: 1/2007-1/2010, Surface (0-20 m) concentrations in August 2008 (left) and January 2010 (right) after continuous release of 1 TBq/y off Murmansk in 10^-2 Bq/m³.

(18)

Figure 8: Concentration of the radionuclide (10^-2 Bq/m³) in surface water (0-20 m) for Experiment B (1/2007 – 1/2010), just offshore of Vardø (top), southern Barents Sea, near 73ºN 30ºE (middle) and at the southeastern tip of Svalbard near Edgeøya (bottom).

(19)

Figure 9: Experiment C: 1/2010-1/2013, Surface (0-20 m) concentrations in December 2011 (left) and January 2013 (right) after continuous release of 1 TBq/y off Murmansk in 10^-2 Bq/m³.

(20)

Figure 10: Concentration of the radionuclide (10^-2 Bq/m³) in surface water (0-20 m) for Experiment C (1/2010 – 1/2013), just offshore off Vardø(top), southern Barents Sea, near 73ºN 30ºE (middle) and at the southeastern tip of Svalbard near Edgeøya (bottom).

(21)

3.1.2 Terrestrial advection-dispersion modelling results

The maximum total deposition to Norway for all model runs covering 33 years of bi-daily runs resulted in a

137Cs and ⁹⁰Sr deposition of 97 and 79 TBq, respectively. The value for ¹³⁷Cs can be compared to total fallout of ¹³⁷Cs over Norway after the Chernobyl accident of 2300 TBq (Henriksen and Saxebøl, 1988). The absolutely worst case with respect to deposition over Norway can be found in one of the model runs (on 1988-07-22 with release between 06-07 UTC), resulting in a deposition where more than half of what has been released ends up in Norway (54%). More on model runs, assumptions made, and the selection of worst cases related to highest deposition in Norway can be found in an earlier report (Hosseini et al. 2017).

3.1.3 Activity concentrations in marine organisms for hypothetical releases

Activity concentrations of ¹³⁷Cs and ⁹⁰Sr in marine organisms, resulting from releases into the marine environment for a seawater circulation pattern corresponding to that occurring in the period January 1998 – January 2001, are shown in Fig. 11.

The activity concentrations of ¹³⁷Cs in marine organisms (Fig. 11) do not exceed 50 Bq/kg fresh weight at any time during the 3-year simulation period. The highest levels are predicted to occur in seabirds, for which ¹³⁷Cs activity concentrations fluctuate between 20 and 50 Bq/kg, once the system has attained a quasi-steady-state. Nonetheless, given the limited empirical datasets underpinning the biokinetic models for seabirds, these estimates should be considered as being quite uncertain. Cs-137 activity

concentrations are relatively low for fish attaining peak levels only slightly in excess of 5 Bq/kg once 280 days have elapsed and oscillating in magnitude in proximity to this value for the remaining simulation period.

An alternative to applying a biokinetic model, as illustrated above, would have been to apply concentration factors, CFs, to seawater activity concentrations as recommended elsewhere (IAEA, 2004). For the

particular case illustrated in Fig. 11, ¹³⁷Cs activity concentrations in seawater were predicted to fluctuate Figure 11: Activity concentrations of ¹³⁷Cs (Bq/kg f.w.) and ⁹⁰Sr (Bq/kg f.w.) in marine organisms with time for a release of 220 TBq ¹³⁷Cs and 180 TBq ⁹⁰Sr occurring over 3 years with oceanic circulation regime for inputs starting in 1998.

(22)

13 Bq/kg. These are predictions that are commensurate with levels determined via the application of a biokinetic model and suggest that the simpler approach, i.e. using CFs, may have sufficed in providing a reasonable indication of food-chain transfer in this case. Nonetheless, we contend that the validity of the biokinetic model is greater and that the smoothing of extreme values, noting the peak for the biokinetic model output for ¹³⁷Cs in fish is < 50 % of that derived via a CF, would be a real, observable phenomenon should a substantial release of ¹³⁷Cs occur.

Despite the hypothetical releases of ⁹⁰Sr into the marine environment being somewhat lower than those associated with ¹³⁷Cs, the activity concentrations of ⁹⁰Sr in most marine organism categories are predicted to be slightly higher than corresponding ¹³⁷Cs levels. In some cases, notably for mammals and birds, this may be partly explained by slower loss rates of strontium compared to caesium once the radionuclide becomes incorporated into an animal’s body. The fact that activity concentrations of ⁹⁰Sr in mammals and birds are still increasing at the end of the simulation period is indicative of the marine system having not attained a steady state (i.e. equilibration between activity concentrations in seawater and some biota compartments has not occurred within the 3-year simulation period).

The relative transfer of ⁹⁰Sr to different organism groups shows a similar pattern to that observed for ¹³⁷Cs with seabirds exhibiting the highest uptake and crustaceans and fish the lowest. The highest activity concentration predicted to occur is in excess of 80 Bq/kg f.w. ⁹⁰Sr in seabirds and all organism groups exhibit ⁹⁰Sr activity concentrations in excess of 3 Bq/kg f.w. once quasi-steady-state conditions have been attained.

Activity concentrations of ¹³⁷Cs and ⁹⁰Sr in marine organisms, resulting from releases into the marine environment for a seawater circulation pattern corresponding to that occurring in the period January 2007 – January 2010, are shown in Fig. 12 and that for the period January 2010 – January 2013, are shown in Figures 13.

Figure 12: Activity concentrations of ¹³⁷Cs (Bq/kg f.w.) and ⁹⁰Sr (Bq/kg f.w.) in marine organisms with time for a release of 220 TBq ¹³⁷Cs and 180 TBq ⁹⁰Sr occurring over 3 years with oceanic circulation regime for inputs starting in 2007.

(23)

The oceanic circulation regime does appear to make some subtle difference to the form of the temporal evolution of radionuclide activity concentrations in biota, but the absolute levels are generally consistent falling within a few Bq/kg of one another for any given time point. The radionuclide activity concentrations in the organism category ‘Mollusc’ appear to exhibit the greatest contrast in levels over time in response to different seawater circulation patterns. Whereas activity concentrations of ¹³⁷Cs in molluscs show peak values occurring at around 200 (simulation) days for the circulation regime starting in 2007, maximum levels occur at around 600 (simulation) days for the circulation regime starting in 2010.

In order to provide some context to the activity concentrations of radionuclides predicted in the marine environment for hypothetical releases from Andreeva Bay, under different oceanic circulation regimes, it is helpful to consider recent levels, in relevant sea areas, based on measurements (Table 1). Reflecting historical inputs from, inter alia, atmospheric nuclear testing and western European reprocessing plans (AMAP, 1998) relatively low, but measurable quantities of ¹³⁷Cs can be determined in seawater in European Arctic seas.

Table 1. Measured activity concentrations of ¹³⁷Cs in seawater and fish based on samples collected in the Barents Sea and Norwegian Sea in recent years.

Location

Reference Water (Bq/l) Biota (Bq/kg f.w.)

Barents Sea 1.6E-03 – 2.0E-03 Fish <0.3 (Gwynn et al.

2012)

Norwegian Sea 1.1E-03 – 5.9E-03 Fish* <0.5 NRPA (2011)

& NRPA (2015)

*Caught within the coastal waters of Finnmark and Troms.

Clearly, the signal from the specified hypothetical release, in terms of ¹³⁷Cs activity concentrations in Figure 13: Activity concentrations of ¹³⁷Cs (Bq/kg f.w.) and ⁹⁰Sr (Bq/kg f.w.) in marine organisms with time for a release of 220 TBq ¹³⁷Cs and 180 TBq ⁹⁰Sr occurring over 3 years with oceanic circulation regime for inputs starting in 2010.

(24)

predicted activity concentrations of ¹³⁷Cs in seawater and marine organisms lie at least one order of magnitude above those currently existing in the considered sea areas. Given that ⁹⁰Sr background activity concentrations in marine organisms from the Barents region have historically been observed to be considerably lower than corresponding levels associated with ¹³⁷Cs (Heldal et al., 2015) and that

radionuclide levels in fish are indicative of ¹³⁷Cs and ⁹⁰Sr in other marine organism groups, it is safe to say that a similar argument, regarding the distinctive measurability of (hypothetical) Andreeva inputs on marine food-chains should there be an accident, holds true.

All activity concentrations in marine organisms, as predicted by models for hypothetical release situations, that could be used as food for humans are substantially below the intervention level of 600 Bq/kg as applied for basic foodstuffs in Norway (Liland et al., 2009). Furthermore, the predicted ¹³⁷Cs activity concentrations in fish, are substantially below the Japanese regulation value of 100 Bq/kg (fresh weight) for sale and human consumption- a limit applied following the accident at Fukushima Daiichi (Buesseler, 2012). Given the highly conservative nature of our assessment and the undoubted highly pessimistic prognosis regarding the contamination of food-chains, the potential requirement for restrictions on the consumption of sea products, should a release during the transport from Andreeva Bay occur, can be summarily dismissed. This contrasts with similar assessments made considering potential releases from the sunken submarine K-159 in the Barents Sea for which Antipov et al. (2015) considered that restrictions, at least in the short term, would plausibly be required.

3.1.4 Activity concentrations in terrestrial food-chains for hypothetical releases

The dynamics of ¹³⁷Cs activity concentrations in vegetation (shrubs), lichen, small mammals and large (mammalian) herbivores at the location of maximum deposition in Finnmark, following releases associated with a hypothetical accident scenario during transport from Andreeva Bay, are presented in Fig. 14.

Figure 14: Activity concentrations (Bq/kg f.w.) of ¹³⁷Cs in selected organisms for the area of maximum deposition in Finnmark for the Andreeva Bay release scenario. The red dashed hortizontal line indicates the current Norwegian intervention level for game/reindeer meat of 3000 Bq/kg.

(25)

For lichen and shrubby vegetation, maximum ¹³⁷Cs activity concentrations are predicted to occur immediately after the hypothesized deposition event with levels declining in the subsequent weeks and months as the radionuclide is weathered from the (above soil-surface) canopies. Initial ¹³⁷Cs activity concentrations were simulated to be in excess of 27 kBq/kg (f.w.) for shrub and in excess of 35 kBq/kg (f.w.) for lichen primarily reflecting the interception fractions used in model parameterization. The decrease in ¹³⁷Cs levels with time follow quite divergent trajectories. Whereas activity concentrations in lichen fall slowly, remaining at levels above 30 kBq/kg (f.w.) by the end of the 1-year simulation period, the activity concentrations in shrub decline by a factor of about 40, so that concentrations are substantially below 1 kBq/kg (f.w.) by the corresponding time point. Both small and large mammals exhibit ¹³⁷Cs activity

concentration maxima that occur at a finite time period post deposition. This maximum occurs at 20 days after deposition for small mammals and at around 170 days for large herbivores. The ¹³⁷Cs activity concentrations in small mammals attain a maximum that is > 50 kBq/kg (f.w.) but this level falls relatively quickly to a value of approximately 3.5 kBq/kg (f.w.) by the end of the simulation period. In contrast, ¹³⁷Cs activity concentrations in large herbivores exhibit absolute maxima that are lower, approximately 24 kBq/kg (f.w.), but decline much more slowly from peak levels and remain substantially in excess of 22 kBq/kg (f.w.) by the end of year 1.

The terrestrial model applied in the current assessment had been developed earlier (Hosseini et al., 2016;

2017 and Brown et al., 2016b). During this earlier process and to reflect the uncertainty associated with using a relatively untested model, a check for ¹³⁷Cs large herbivore predictions was made against a more established model developed specifically to look at radiocaesium levels in reindeer following Chernobyl accident derived deposition (see Åhman, 2007). The previous study found some reassuring similarities in terms of the maximum activity concentrations that would be predicted for similar ¹³⁷Cs deposition inputs to the models. Nonetheless, it was acknowledged that the Åhman model provides an additional advantage by providing a quantification of activity levels in the years following a deposition event. It was further noted by Brown et al. (2016b) that an interesting feature of Åhman’s model is the closeness with which the observed high amplitude, approximately sinusoidal form of ¹³⁷Cs levels in reindeer can be simulated by accounting for factors such as the animal’s changing diet and metabolism over the season. From the earlier analysis it was concluded that new peak activities, coinciding with subsequent winter periods, would be expected and that these subsequent peaks might be expected to fall at levels that were not

dramatically lower than the initial maximum. In view of our using the same model for this analysis, similar inferences can be invoked for the current assessment.

The activity levels for small (burrowing) mammals and deer can be compared to the current relevant Norwegian intervention level of 3000 Bq/kg f.w for radiocaesium (as applied following the Chernobyl accident) and applicable to, inter alia, reindeer and game (Thørring et al., 2010). Clearly, ¹³⁷Cs activity concentrations in large herbivores (nominally reindeer) are substantially in excess of the intervention level for the virtually the entire simulation period. For small mammals/game, ¹³⁷Cs levels are much higher than the intervention level in the initial period, i.e. first 100 days or so of the assessment, falling to levels marginally above the intervention level towards the end of the first year. The introduction of restrictions on the consumption of reindeer/game would, therefore, be highly likely, potentially involving various measures such as the dissemination of dietary advice and, for reindeer at least, the use of live monitoring before slaughtering. Although the simulations suggest a fairly rapid decline in ¹³⁷Cs activity concentrations within the first year, the uncertainty associated with the longer-term predictions would precipitate the need for careful monitoring of the situation. In other words, model prognoses of this type could not be used in isolation to predict for how long any introduced restrictions would need to be retained. Similarly, for shrubs (used as a proxy for berries) restrictions on fruit consumption might foreseeably need to be introduced, at least in the first harvesting season. The highest predicted levels are substantially above the

(26)

The dynamics of ⁹⁰Sr activity concentrations in vegetation (shrubs), lichen, small mammals and large (mammalian) herbivores at the location of maximum deposition in Finnmark, following releases associated with a hypothetical accident scenario during transport from Andreeva Bay, are presented in Fig. 15.

The temporal profiles representing ⁹⁰Sr activity concentrations in lichen and vegetation/shrub are similar in form to those observed for ¹³⁷Cs. Both lichen and shrub exhibit maxima, of ca. 33 kBq/kg (f.w.) ⁹⁰Sr for shrub and 15 kBq/kg (f.w.) ⁹⁰Sr for lichen, at the start of the simulation but ⁹⁰Sr levels in lichen decline relatively slowly whereas vegetation levels are predicted to fall relatively quickly attaining levels that are 260 times lower than the start point once 365 days have elapsed (cf. corresponding decrease for ¹³⁷Cs was ca. 40). The temporal profiles for ⁹⁰Sr levels in mammals contrast quite distinctly with ¹³⁷Cs activity

concentrations in mammals vs time. The maximum ⁹⁰Sr of 49 kBq/kg (f.w.) in small mammal is predicted to occur at a time approximately 80 days after deposition, i.e. somewhat later than for radiocaesium, but in contrast to the rapid falloff seen for ¹³⁷Cs, activity concentrations decline relatively slowly and are still at a level exceeding 37 kBq/kg ⁹⁰Sr after 365 days of elapsed time. For large herbivores, no steady state appears to have occurred between environmental compartments even following 1 year of simulation as levels of ⁹⁰Sr are still increasing at this stage. The maximum level at the end of the simulation in large herbivore is approximately 18.5 kBq/kg ⁹⁰Sr (f.w.). As noted earlier in Hosseini et al. (2016) when applying an earlier version of the aforementioned biokinetic model, there is an expectation that ⁹⁰Sr will be lost

relatively slowly from the body of mammals following assimilation because the radionuclide rapidly becomes associated with bone (see Coughtrey & Thorne, 1983). Once sequestered by bone the retention of radiostrontium within the animal’s body can be protracted, reflecting the skeletal turnover of analogous elements such as calcium. This is congruent with the model predicting a continued increasing trend in ⁹⁰Sr activity concentrations in reindeer towards the end of the simulation.

Nonetheless, Hosseini et al. (2016) also considered the observation of no equilibrium (in ⁹⁰Sr) being attained over the 1-year simulation period could simply reflect poor parameterisation of the model. For example, in the case of reindeer, the diet and the biological half-lives (for radiocaesium and by proxy

Figure 15. Activity concentrations (Bq/kg f.w.) of ⁹⁰Sr in selected organisms for the area of maximum deposition in Finnmark for the Andreeva Bay release scenario.

(27)

considerations were not accounted for in the model. Despite the acknowledgment that the configuration of the model is limited after the first few months, it is not unrealistic to expect a continued build-up of ⁹⁰Sr during later periods. This is based on the assumption that ⁹⁰Sr is strongly retained on lichen thus leading to a relatively slow decline in levels with time as suggested by Golikov et al. (2004) and the knowledge that lichen do form a large part of the reindeer’s diet at some periods of the year. Under such conditions, considerable time would be required to attain anything approaching equilibrium. Having said this, the appropriateness of using an approximately 5 year weathering half-time value for ⁹⁰Sr may be questionable in view of evidence to the contrary where (more or less corresponding) effective half-times of about 1.0 year for ⁹⁰Sr in lichen have been apportioned (Skuterud et al., 2005). Although the value of 5 years has been retained for the sake of conservatism, the effect of applying a lower weathering half-time for ⁹⁰Sr in lichen would be to observe a more rapid attainment of a maximum activity concentrations in reindeer.

3.2 Doses to humans

3.2.1 Doses to humans from marine exposure pathways

Annual effective doses to people representing critical groups from the Kola Peninsula and represented by a high rate seafood consumer in Norway are shown in Table 2.

Table 2. Estimated annual committed effective doses to human for ingestion of marine foodstuffs based on a hypothetical release at Andreeva Bay.

Oceanic Circulation Pattern - period

Annual effective dose

(µSv) Critical Group - Comments

1998-2001 13

Representative person in identified population group in Kola Peninsula, based on

ingestion of fish, molluscs and crustaceans

2007-2010 11

2010-2013 13

1998-2001 37

Representative person in identified population group in Norway - high rate

consumers of seafood

2007-2010 31

2010-2013 39

The annual doses fall in the range between 11 to 39 µSv. The doses attributable to high rate consumers of seafood in Norway are approximately a factor of 3 greater than critical groups from the Kola Peninsula, simply reflecting the much greater consumption of seafood in the former group.

These doses may be considered to be low. The radiation dose to the average Norwegian from food and drinking water has been estimated to 0.52 mSv/year from naturally occurring sources (Komperød et al., 2015). Virtually the entire 0.19 mSv/year attributable to the ingestion of fish and shellfish containing

(28)

current doses attributable to the ingestion of seafood containing natural ‘background’ radioactivity for the Norwegian population is approximately a factor of 10 greater than the corresponding hypothetical doses that could arise from releases of artificial radionuclides to Andreeva Bay. The Norwegian population undoubtedly lies at the higher end of statistical compilations for the consumption of seafood across the globe. As considered by Aarkrog et al. (1997), the global mean individual dose from ²¹⁰Po in seafood (6 µSv) corresponds to about 1 day’s effective dose from all natural sources of radioactivity (2.4mSv per year according to UNSCEAR (1993)). So, this contribution from ²¹⁰Po in seafood alone (noting that this value would undoubtedly be higher for high consumers of seafood in Norway) would be commensurable with the annual dose predictions being made for the hypothetical accident for the marine exposure pathway.

Similarly, reference can be made to various dose benchmarks typically considered within a regulatory context. As noted by Hosseini et al. (2017), in the context of setting criteria to allow or forbid the dumping of radioactive material at sea, the IAEA (1999, 2015) have used a de minimis dose of 10 µSv per annum. This effective dose (in tandem with a defined collective effective dose) to a representative person would allow a practice to be exempted from further regulatory consideration. The de minimis level specified above provides a robust indication of what might be widely considered to be a trivial radiation dose and as such provides an appropriate benchmark with which the doses calculated in this assessment might be

compared. The annual doses calculated for the hypothetical release from the marine exposure pathway, therefore, fall only slightly above a level that is putatively regarded as trivial.

It goes without saying that, the doses fall within the ICRP’s reference band, germane to accidental releases of radioactivity, ‘below 1 mSv’ where a requirement for significant intervention would not be deemed appropriate (ICRP, 2007).

3.2.2 Doses to humans from terrestrial exposure pathways

Annual effective doses to people representing Norwegian critical groups living in a region of maximal deposition (i.e. within Finnmark) are shown in Table 3. It should be noted that the focus of the assessment has been on Norway with no effort placed in assessing doses for locations in greater proximity to the hypothetical release point.

The annual effective dose to a representative person in Finnmark exceeds 20 mSv per year and would be dominated by the ingestion pathway. The finding that the ingestion pathway would dominate the dose in Finnmark is congruent with earlier analyses for submarines K-27 (Hosseini et al., 2016) and K-159 (Hosseini et al., 2017) with scenarios following a salvaging operation and fires. This, however, is not a corroboration of the analysis because the same approach was used in all of these assessments, i.e. the same atmospheric advection dispersion models, food-chain transfer models and dose calculation methodologies were applied.

It should be further noted that the 20 mSv calculated assumes no intervention, with the continued ingestion of the most highly contaminated products over a 1-year period. Self-evidently, in the event of a real accident with releases at the (worst case) hypothesized levels, food interventions levels would be exceeded (see discussion above) and critical groups would be controlled from ingesting highly contaminated foodstuffs.

(29)

Table 3. Estimated annual effective doses to individuals in identified population group in Norway for various pathways based on releases to atmosphere at Andreeva Bay.

Pathway Annual effective dose (mSv)

Inhalation 1.36E-02

Cloud shine 4.00E-05

Ground shine 4.60E-01

Ingestion (terrestrial food) 2.07E+01

Total 2.12E+01

Internal doses from terrestrial food products are almost three orders of magnitude greater than the internal doses associated with a high rate fish consumer living in Norway. This clearly illustrates that the overwhelmingly predominant risk to Norwegian populations should an accident involving the transport of SNF from Andreeva to Murmansk would be via an atmospheric transport pathways and terrestrial food- chains.

The total average dose from ionising radiation to the Norwegian population (including medical applications) has been estimated to 5.2 mSv/year with the single highest dose contribution caused by inhalation of radon in indoor air (Komperød, 2015). The hypothetical doses derived in the current analysis are a factor of 4 greater than this value and might therefore be considered substantial.

The ICRP (2007) set a Reference level band of 20 to 100 mSv set for the highest planned residual dose from a radiological emergency and the calculated values within the current assessment fall just into the lower end of the specified range. If doses fall within such a band the ICRP note that consideration should be given to reducing doses, Individuals should receive information on radiation risk and on the actions to reduce doses and assessment of individual doses should be undertaken. In fact, for Norway, if an accident did occur and projected doses were as high as those considered above, reference to prescribed protective measures for the public such as those published in relevant guidance (Nordic Flagbook, 2014) would be required. It is noted in the Nordic “Flagbook” that If the projected dose due to the radiation hazard without any protective measures in higher than 10 mSv then it would be necessary to apply appropriate protective measures.

3.3 Doses to biota

3.3.1 Dose rates in marine organisms

Dose rates in marine organisms for the 3 specified oceanic circulation patterns are shown in Fig. 16. The organisms can be essentially split into 2 groups – pelagic and benthic. The pelagic group, comprising of (pelagic) fish, mammals and seabirds, exhibit relatively monotonous dose-rates over time (following the initial increase) reflecting the fact that a predominant proportion of their dose rate is derived from

internally-incorporated radionuclides. In contrast, dose-rates for the benthic group, comprising of molluscs and crustaceans, exhibit dose-rates that fluctuate substantially over time. In this case, external exposure

(30)

organism’s habitat, i.e. at the sediment-water interface. Sediment activity concentrations of ¹³⁷Cs and ⁹⁰Sr were derived using a kd, for which instantaneous equilibration between activity levels in seawater and sediment (at the seabed) are assumed. Not only would radionuclide activity concentrations in sediment not be expected to attain steady state conditions with radionuclide levels in the overlying water column so quickly (i.e. non-instantaneously), there is, arguably, an additional conservatism that has been introduced by using activity concentrations in surface waters. For a surface release of radionuclides, as modelled in this work, radionuclide levels in seawater in contact with the seabed might be expected to exhibit substantially lower levels than those observed at the surface.

The highest dose-rates are observed for the benthic group, with crustaceans and molluscs exhibiting dose rates often exceeding 0.1 μGy/h over the course of the simulation and, in the most extreme case, attaining exposure levels exceeding 0.25 μGy/h for a short period in the case of the circulation pattern from 2010.

The levels of exposure of pelagic organisms are substantially lower, by contrast, and only slightly exceed 0.05 μGy/h for a short period of time in the case of seabirds and the circulation patterns for 1998 and 2010. Dose rates for other pelagic organisms are relatively low for all circulation patterns, with fish exhibiting exposures substantially below 0.01 μGy/h.

The effect of circulation pattern is quite subtle. All 3 current regimes, starting in 1998, 2007 and 2010, exhibit 3 distinct maxima in the benthic organism’s dose rates and these occur approximately during the same time periods. It is only really in the absolute levels of exposure that differences become more

evident. The highest dose-rates for benthic organisms for the 1998 circulation pattern are observed during the 1^st and 3^rd periods of elevated exposure and for which levels are approximately commensurate. In contrast, the highest exposure for the 2007 circulation are observed during the 1^st peak and that for the 2010 circulation for the 2^nd peak.

There are two common ways to contextualize dose-rates to wild organisms (Brown et al., 2004a). One is to compare the dose rates to natural background radiation, arising from the presence of primordial (and sometimes cosmogenic) radionuclides in the environment and the other is to compare against benchmarks that inform us about the levels at, or above, which some form of deleterious effect is likely to occur.

Weighted (to account for differences in the magnitude of deleterious effects associated with radiation quality) absorbed dose rates from exposure to naturally occurring radionuclides of 5.8 × 10⁻¹ and 8.7 × 10⁻¹ µGy/h were calculated by Hosseini et al. (2010) for a set of marine organisms comprising of flatfish, crabs

Figure 16. Dose-rate (μGy/h) for fish, molluscs, crustacea, mammals and seabirds for the oceanic circulation pattern of 1998 (left), for the oceanic circulation pattern of 2007 (middle) and for the oceanic circulation pattern of 2010 (right).

(31)

great differences that are often observed in the activity concentrations of naturally-occurring

radionuclides in natural systems. For benthic organisms, exposures will be particularly associated with variability in radionuclide activity concentrations in sediments, generally reﬂecting, in turn, the levels in the source rock from which they were derived, and furthermore the grain size. The maximum dose rates

calculated in the present study are substantially lower than the aforementioned dose rates attributed to natural background radiation in the marine environment albeit falling within the same order of magnitude.

Most of the dose rates calculated, and most notably for fish, mammals and seabirds, are at least one order of magnitude below the dose rates associated with naturally occurring radionuclides. For this group of pelagic organisms, one can conjecture that the projected dose rates would fall within the variability associated with the exposure of marine organisms, in general, to natural background radiation over a wide geographical region.

With regards to benchmarks, the ICRP (ICRP, 2008) recommend the application of a set of Derived

Consideration Reference Levels (DCRLs) for particular categories of Reference Animals and Plants (RAPs) (Fig. 17). These are defined as “A band of dose rate within which there is likely to be some chance of deleterious effects of ionising radiation occurring to individuals of that type of Reference Animal or Plant (derived from a knowledge of defined expected biological effects for that type of organism) that, when considered together with other relevant information, can be used as a point of reference to optimise the level of effort expended on environmental protection, dependent upon the overall management objectives and the relevant exposure situation.”

Although dose-rates for the marine RAPs included in ICRP (2008) have not been calculated explicitly, apart from crustacean/crab, it should be fairly clear that all of calculated dose-rates fall at least two orders of magnitude below the lower end of the DCRLs associated with marine RAPs – 1 mGy/d or 42µGy/h. As such, the dose rates derived in the present study would be far below the threshold above which one might

Figure 17. Derived Consideration Reference Levels (DCRLs) for environmental protection for each Reference Animal or Plant (RAP), the RAPs being grouped according to their terrestrial, freshwater, or marine habitat. Reproduced from ICRP (2014).

(32)

3.3.2 Dose rates in terrestrial organisms

Absorbed dose rates vs time in terrestrial organisms located in the maximum deposition area in Finnmark following an atmospheric release from Andreeva in line with the specification stipulated earlier is presented in Fig. 18.

The absorbed dose rate in vegetation shrub exceed 20 µGy/h at the start of the simulation but decrease rapidly thereafter and fall substantially below 1 µGy/h once 50 days have elapsed. In contrast both small and large mammals exhibit dose rates that increase substantially following the initial deposition event reflecting the time lag for radionuclides to be transferred through the food-chain. Maximum dose rates, exceeding 40 µGy/h, are observed for small mammals within the first 50 days of simulation whereas maximum dose rates for large mammals, exceeding 20 µGy/h, are observed at the end of the simulation period. The fact that dose-rates for large mammals are still increasing at time = 365 days reflects the observation that ⁹⁰Sr levels have not attained a steady state between environmental compartments as discussed previously. A summarised table for dose rates in terrestrial biota is provided in Table 4.

The calculated dose-rates for all organisms exceed the ERICA screening benchmark of 10 µGy/h at various stages during the course of the simulation. Although strictly applicable to long term planned releases of radionuclides to the environment, the exceedance of this benchmark would lead to a recommendation to provide a more detailed analysis of the exposure of non-human biota. In other words, the predication of environmental dose rates greater than 10 µGy/h dose rate not necessarily indicate any detrimental impact but would be at such a level where the absence of deleterious effect cannot be guaranteed.

Figure 18. Dose-rate (μGy/h) for terrestrial biota – small mammal/game, large (mammalian) herbivore and Vegetation/shrub – in Finnmark following an atmospheric release from Andreeva. The red dashed horizontal line indicates the ERICA screening benchmark (Brown et al., 2016c).